Wrap-up "Machine learning" by Andrew Ng

I attended a online course, "Machine learning" by Andrew Ng that held in Coursera.org by Stanford Univ. Online.

I started this course to find more background theories about "Tensorflow".

However, it makes me be more interested in "Machine learning" and more curious about "Deep learning".



# Some impressive things in the course

- The professor explains the detailed algorithms with many examples.

   If I didn't understood an algorithm, I just waited and kept seeing the examples. The examples helped me understand.

- He explains the mathematical approaches of the algorithms.

   So I had to stop the video and tried to the mathematical approaches by myself.

   It took more times than I planned.

- The exercises are so difficult to make vectorized codes.

   But I think it is the most important point to imagine inner algorithms of "Tensorflow".

   I needed to practice matrix calculations in Octave to imagine vectorized codes.

- He knows the machine learning algorithm and system are difficult to understand and he said he also had felt them difficult when he learn.

   This words encouraged me.



# Wrap-up

1. Supervised Learning

    a. Linear regression

    b. Logistic regression

    c. Neural networks

    d. SVMs(Support Vector Machine)


2. Unsupervised Learning

    a. K-means

    b. PCA(Principle Component Analysis)

    c. Anomaly detection


3. Special applications / special topics

    a. Recommender systems

    b. Large scale machine learning


4. Advice on building a machine learning system

    a. Bias / Variance

    b. Regularization

    c. Deciding what to work on next

        - Evaluation of learning algoritms

        - Learning curves

        - Error analysis

        - Ceiling analysis

'Machine Learning' 카테고리의 다른 글

YOLO v3 demo webcam 돌려보기 for ubuntu(18.04)  (0) 2018.10.03
Activation functions  (0) 2017.11.30
Binomial Logistic Regression/Classification  (0) 2017.07.10
Linear Regression  (0) 2017.07.07

+ Recent posts